sign in sign up
<<
Home , Cochlear nucleus, Dorsal cochlear nucleus, Hair cell, Spiral ganglion, Superior olivary complex, Ventral cochlear nucleus

THE JOURNAL OF COMPARATIVE NEUROLOGY 278:591-603 (1988)

Brainstem Branches From Olivocochlear in Cats and Rodents

M.C. BROWN, M.C. LIBERMAN, T.E. BENSON, AND D.K. RYUGO Departments of Physiology and Otolaryngology (M.C.B., M.C.L.), and Department of Anatomy and Cellular Biology (T.E.B., D.K.R.), Harvard Medical School, Boston, Massachusetts 02115; Center for Sciences, Johns Hopkins University School of Medicine, Baltimore, Maryland 21205 (D.K.R.); Eaton-Peabody Laboratory, Massachusetts Eye and Infirmary, Boston, Massachusetts 02114 (M.C.B., M.C.L., T.E.B., D.K.R.)

ABSTRACT Horseradish peroxidase was used to label axons of olivocochlear (OC) by intracellular injections in cats and extracellular injections in rodents. These axons arise from cell bodies in the and project to the . En route to the cochlea, the thick axons (> 0.7 pm diam.) of medial olivocochlear (MOC) neurons formed collaterals that terminated in the ventral , the interstitial nucleus of the vestibular (in cats), and the inferior vestibular nucleus (in rodents). The thin axons (< 0.7 pm diam.), presumed to arise from lateral olivococh- lear (LOC) neurons, did not branch near the CN. Within the CN, the MOC collaterals tended to ramify in and near regions with high densities of granule cells, regions also associated with the terminals of type I1 afferent sons (Brown et al.: J. Comp. NeuroL 278.581-590, '88). These results suggest that those fibers associated peripherally with outer hair cells (MOC efferents and type I1 afferents) are associated centrally with regions contain- ing granule cells, whereas those fibers associated with inner hair cells peripherally (LOC efferents and type I afferents) are not.

Key words: cochlear nucleus , vestibular nucleus, , collateral, efferent nerve

Rasmussen ('46, '53) published the initial descriptions of The present study concentrates on the morphology of the the olivocochlear bundle (OCB) from its sources in the su- OC collaterals within the . These collaterals have perior olivary complex to its presumed endings in the coch- not been described systematically, although they are known lea. More recent work (Warr and Guinan, '79; Guinan et to innervate the WCN; Rasmus- al., '83; White and Warr, '83) has divided the olivocochlear sen, '60, '67; Liberman and Brown, '86). We used horserad- (OC) neurons into two systems according to location ish peroxidase (HRP) to label collaterals of OC neurons: with respect to the medial superior olive: the medial olivo- intracellular injections for physiologically characterized cochlear (MOC) system and the lateral olivocochlear (LOG) single neurons, and focal extracellular injections for small system. Peripherally, the thick myelinated axons of the populations of neurons. Both types of injections can also MOC neurons terminate mainly on outer hair cells, whereas label primary afferent neurons in the same tissue (Liber- the thin and apparently unmyelinated axons of LOC neu- man and Brown, '86; Brown et al., '88a). Thus, the methods rons terminate predominantly on afferent just permit us to directly compare the trajectories and termina- beneath inner hair cells (Liberman, '80; Ginzberg and Mo- tions of afferent and efferent fibers in the CN. These de- rest, '83; Guinan et al., '83; Liberman and Brown, '86; scriptions are relevant to an understanding of the Brown, '87). The anatomical differences between the MOC relationship of the feedback system to the afferent and LOC systems suggest at least two general types of pathways. feedback control of the periphery. At present, however, there is no consensus as to the function of either OC system in the awake, behaving animal. Further studies of the anatomical and physiological characteristics of the efferent systems may offer clues to their functional significance. Accepted June 29,1988.

0 1988 ALAN R. LISS, INC. M.C. BROWN ET AL. MATERIALS AND METHODS RESULTS Intracellular injections Classification and general brainstem course of OC Single OC units were recorded in the internal auditory fibers meatus from the region near the vestibulocochlear anasto- In seven cats, nine OC fibers were labeled following intra- mosis in seven Dial-anesthetized (0.73 mag, i.p.) cats as cellular injection near the vestibulocochlear anastomosis. described previously (Liberman and Brown, '86). At the Labeled cell bodies were found in brainstem loci that have time of recording, fibers were classified as "efferent" on the been described as the origins of the medial olivocochlear basis of 1) position within the nerve bundles, 2) regularity (MOC) cell group (Warr and Guinan, '79; Guinan et al., '83). of the interspike intervals, and 3) latency of responses (5- Single OC units respond to monaural sounds presented to 50 msec) to tone bursts (Liberman and Brown, '86). Neurons either the ipsilateral or the contralateral or to both were labeled by iontophoresis of HRP, and injections were and have tuning curves with a well-defined characteristic limited to one or two fibers per animal. After a survival frequency (Liberman and Brown, '86). The units labeled in time of about 30 hours, the anesthetized cats were perfused this study ranged in characteristic frequency from 1.9 to 21 intracardially with glutaraldehyde and paraformaldehyde. kHz; six responded only to ipsilateral sound, and three The cochlear nucleus was either separated from the - responded only to contralateral sound. stem and processed with diaminobenzidine (Fekete et al., The MOC axons course from their cell bodies (Fig. 1)in+ a '84) or was left attached to the remainder of the brainstem, dorsal direction, approaching the surface of the brainstem sectioned on a freezing microtome, and processed with te- just beneath the . As the axons travel lat- tramethylbenzidine (Mesulam, '82). Axons in material pro- erally after approaching the midline, they follow the vestib- cessed with tetramethylbenzidine could be traced for long ular nerve root and run ventromedial to the anteroventral distances and their branching patterns could be recon- cochlear nucleus (AVCN). Within the root, structed, but the crystalline nature of the label prevented MOC axons produce collaterals. The MOC axons cross the detailed analysis of fiber or terminal morphology. For fine inferior vestibular , enter the auditory nerve (at resolution of the terminal regions, material processed with the vestibulocochlear anastomosis), and terminate periph- diaminobenzidine was used. erally on outer hair cells (Liberman and Brown, '86). The MOC axons are thick (about 2-4 pm diameter in the vestib- Extracellular injections ular nerve root) and have periodic constrictions that we Extracellular injections of HRP were made into the spiral interpret as nodes of Ranvier (Fekete et al., '84; Liberman ganglion region as described previously (Brown et al., '88a). and Oliver, '84). Although thin unmyelinated efferent ax- Data described are from five gerbils (Meriones unguiculu- ons are also present at the recording site in the vestibulo- tus) and eight CD-1 mice (Mus musculus). All injections cochlear anastomosis (Arnesen and Osen, '841, they have labeled both efferent and afferent fibers. Rodent material not been labeled intracellularly, probably because they are was processed with diaminobenzidine. too thin to be impaled with micropipette electrodes. Pre- Electron microscopy sumably, these thin axons are from LOC neurons and ter- minate peripherally near inner hair cells (Brown, '87). Electron microscopic observations were made from brain- In rodents, extracellular injections of HRP made into the stems in two mice. After a brief intracardiac saline rinse, labeled both thin and thick efferent axons, fixation was achieved by perfusion of a mixture of 0.5% as well as afferent axons. The reaction product within the paraformaldehyde and 1% glutaraldehyde in 0.1 M cacodyl- efferent axons typically faded before reaching the cell body. ate buffer (pH 7.3), followed by 0.5% paraformaldehyde and These partially reconstructed fibers were identified as effer- 3% glutaraldehyde in cacodylate buffer. After 1-hour im- ent because their course was similar to the course of fully mersion in the stronger fixative, the brainstem was dis- reconstructed efferents labeled by intracellular injections. sected and stored overnight in the 0.1 M cacodylate buffer. Medially, the efferent fibers could be retrogradely traced as The tissue was sectioned at 50 pm on a Vibratome. Histo- far centrally as the ventral boundary of the vestibular logical processing for HRP was as described by Fekete et nuclei (Fig. 2, panels 5-71. Laterally, they follow the vestib- al. ('84) with the addition of 1%dimethylsulfoxide to all ular nerve root, which forms the ventromedial boundary of solutions containing cobalt or diaminobenzidine. The sec- the VCN, and exit the brain in the vestibular nerve. Near tions were then stained en bloc with osmium tetroxide and the inferior , the efferent fibers cross to uranyl acetate, dehydrated, and embedded in Epon. La- the auditory nerve at the vestibulocochlear anastomosis. In beled fibers and reference marks such as blood vessels were the cochlea, they cross the spiral ganglion and enter the traced from the Epon-embedded tissue sections at the light intraganglionic spiral bundle before disappearing into the microscopic level with the aid of a drawing tube. Selected dense accumulation of reaction product at the injection site. areas containing drawn fibers were cut from the sections The collaterals given off to the inferior vestibular nucleus and mounted on Epon blanks for ultrathin sectioning and (Fig. 2, panels 3, 4) and to the CN (Fig. 2, panels 1, 2) are electron microscopy. described below. Afferent fibers, in contrast, are restricted to the auditory nerve, bifurcate after entering the CN, and always end within the CN norente de NO, '81; Fekete et Acetylcholinesterase stains al., '84; Fig. 5 in Brown et al., %a). The afferent fibers A modification of the Koelle and Friedenwald method labeled in the present study can often be traced from their was used as described by Osen et al. ('84). Cochlear nuclei cell bodies in the spiral ganglion. were embedded in gelatin-albumin, serially sectioned at 80 The efferent fibers labeled in extracellular injections fall pm in either the coronal or modified sagittal plane, and into two distinct populations based on their diameters (Fig. processed as free-floating sections. 3):thick fibers (diameter > 0.7 pm), and thin fibers (diam- BRAINSTEM BRANCHES OF OLIVOCOCHLEAR AXONS 593

A bbreuiations AN auditory nerve MSO medial superior olive ANR auditory nerve root MVN medial vestibular nucleus AVCN anteroventral cochlear nucleus N. V nucleus of tr. V Cb oc olivocochlear DCN OCB olivocochlear bundle ICP inferior cerebellar peduncle OHC outer hair cell IHC inner hair cell PVCN posteroventral cochlear nucleus IVG inferior vestibular ganglion SCP superior cerebellar peduncle IVN inferior vestibular nucleus Tr. v spinal tract of Vth cranial nerve LC VCN ventral cochlear nucleus LSO lateral superior olive VNR vestibular nerve root LOC lateral olivocochlear VNTB ventral nucleus of the LVN lateral vestibular nucleus VN vestibular nerve or nuclei MCP middle cerebellar peduncle V Vth nerve root MOC medial olivocochlear VIN VIth nucleus MS medial sheet VII VIIth nerve root VIII n. VIIIth nerve

IV'h VENTRICLE

DEEP +-- COCHLEAR NUCLEUS

Fig. 1. Tracing the brainstem course and cochlear-nucleus branching era1 ear stimulation, and the characteristic frequency of its tuning curve pattern of an MOC efferent axon labeled by an intracellular injection of was 4.5 kHz. Inset: Camera-lucida drawing of the labeled soma at high HRP in a cat. The injection site was in the "ipsilateral" vestibulocochlear magnification. The figure has been reversed to make this illustration of an anastomosis (not shown) and the innervated OHCs in the ipsilateral injection into the right ear similar to the succeeding figures, which are of ear. For monaural sound presentations, the unit responded only to ipsilat- injections into left ears. eter < 0.7 pm). Throughout their course, from the injection in the electron micrographs (Fig. 4). The exceptions con- site in the spiral ganglion to the region of the vestibular sisted of one thick efferent for which a profile could not be nuclei (where they usually fade), thick and thin efferent found in the electron micrographs and two faintly labeled fibers are intermingled but retain their differences in di- thin axons found in the electron micrographs that were not ameter. On the basis of light microscopic criteria (e.g., pres- seen in the light microscope. A total of 46 thick fibers and ence of periodic constrictions that label darkly), the thick nine thin fibers were verified to be myelinated and unmy- fibers appear to be myelinated (Fekete et al., '84) and thus elinated, respectively. An analogous correspondence has probably correspond to MOC axons, which arise from cells been shown for auditory primary afferent fibers (Ryugo et in the medial periolivary region. Only two thick fibers in al., '86). one mouse could be traced as far as somata in the ventral One of the cases examined in the electron microscope nucleus of the trapezoid body, a nucleus shown to be the contained two thick axons (about I-pm diam.) that at the major origin of MOC neurons in rodents (White and Warr, light microscopic level did not show the periodic constric- '83). In contrast, the thin fibers have numerous en passant tions characteristic of nodes and were labeled uniformly varicosities, lack obvious constrictions (nodes of Ranvier), throughout their course. These axons proved to be unmy- and label uniformly throughout their length. Most thin elinated at the electron microscopic level. They were the efferent fibers probably originate from LOC neurons only such examples in our light microscopic examination of (Guinan et al., '83; Brown, '87); however, we could not hundreds of OC fibers, and their origin is unclear. verify this, since the reaction product in these fibers faded before reaching the cell bodies. Branching patterns of OC fibers In two mice, examples of thick and thin fibers that had In mice and gerbils, individual thin OC axons were se- been analyzed first in thick sections at the light microscopic lected from the cluster of fibers labeled in the extracellular level were resectioned for examination in the electron mi- injections and followed centrally (Fig. 5). A total of 14 thin croscope to verify the presence or absence of . For axons in four mice and one thin axon in a Ygerbil could be almost all fibers, it was Dossible to match single fibers reconstructed centrally from the saccular ganglion beyond drawn with the light microscope with single IabeleYd profiles the CN before they faded. Two of these thin axons in mice 594 M.C. BROWN ET AL.

1

ventral

poat. .--*. i '.... ,-- , I .* . ! 9.

I I------____.8 Fig. 2. The course of OC fibers (solid lines) labeled in an extracellular upper left panel. Numbers indicated on the coronal view refer to the posi- injection in a mouse, shown schematically in a coronal view (upper left), tions of the modified sagittal sketches. Sections illustrated are 0.24 mm and as actual camera-lucida sketches (1-7) from the tissue sections. The apart. tissue setions were cut in a "modified sagittal" plane as indicated in the BRAINSTEM BRANCHES OF OLIVOCOCHLEAR AXOKIS 595 nerve root (three out of four axons, three of 13 branches), as shown for the neuron in Figure 1. Although the “deep” branch in Figure 1 could not be traced to any of its termi- nals, collaterals from other fibers clearly terminated in the interstitial nucleus of the vestibular nerve, as defined by Brodal and Pompeiano (’57). Axonal branching to both the CN and the interstitial nucleus of vestibular nerve was found for units with low or high characteristic frequencies and for units responsive to ipsilateral or contralateral sound. In mice and gerbils, the number of collateral branches produced per axon is fewer than in cats and their regional distribution is also somewhat different (Fig. 7). Whereas f collaterals in cat innervated the superficial granule-cell layer, branches to this region in rodents were very rare. In addition, the “deep” branches in cat ended in the intersti- tial nucleus of the vestibular nerve, whereas those in ro- dents ended in the inferior vestibular nucleus. In one gerbil, FIBER DIAMETER (pm) seven axons were completely reconstructed’ from the sac- cular ganglion to the ; these axons pro- Fig. 3. Axonal diameters for 37 OC fibers that were labeled in a single mouse, and measured in the vestibular nerve root near the vestibular duced a total of 12 branches (one or two produced per axon). nuclei. Diameters were averaged over segments 10 pm in length and were The branches were directed toward the medial sheet of measured from camera-lucida drawings using a light microscope. granule cells in rostral AVCN (three of 12 branches), the medial sheet in the region of the auditory-nerve root (two could be reconstructed centrally past the vestibular nuclei. of 12 branches), and the inferior vestibular nucleus (seven None of the 15 thin fibers formed collaterals in the brain- of 12 branches). Eight axons were completely reconstructed stem. from one injection in a mouse, producing a total of I1 In contrast, every thick efferent axon, in cats (n = 4) or branches, one or two produced per axon. The branches were in rodents (n = 15), that could be reconstructed from the directed toward the lamina of granule cells dividing PVCN saccular ganglion to the vestibular nuclei formed one or from DCN (three of 11 branches), the medial sheet of gran- more brainstem branches (Table 1).Additionally, when ef- ule cells near the region of the auditory-nerve root (two of ferent branches in the CN were traced back to their source 11 branches), or the inferior vestibular nucleus (six of 11 in the OCB (a total of 101 branches), their parent fibers branches). Some axons formed collaterals only to the were always thick. The OC branches end within the co- cochlear nucleus-others only to the inferior vestibular nu- chlear or vestibular nuclei, with the few OC axons tracea- cleus; when there were two branches, both nuclei were ble to somata showing no further collaterals. The branches innervated (Fig. 5). that end within the CN usually terminate in or near areas OC collaterals formed in the vicinity of the vestibular having high densities of granule cells, which lie along the nuclei in rodents were contined to the vestibular-nerve root lateral and medial edges of the nucleus (Fig. 6).Of the six region near the inferior vestibular nucleus and to the infe- granule-cell regions’ described by Mugnaini et al. (’80a), rior nucleus itself. This nucleus extends posteriorly from OC branches have been found associated with those encom- the vestibular nerve root and contains small and medium- passing the VCN: 1) the superficial layer of granule cells sized cells amongst the descending branches of the vestibu- over the surface of the VCN, 2) the lamina of granule cells lar afferent fibers. OC collaterals terminated only in the dividing the dorsal cochlear nucleus (DCN) from the pos- rostral portion of the inferior nucleus (Fig. 2, panels 3, 4) teroventral cochlear nucleus (PVCN), 3) the medial sheet of and ran parallel to the vestibular fibers before terminating granule cells dividing the VCN from the underlying vestib- in , sometimes near small cells. ular nerve root and brainstem, and 4) the subpeduncular The finding of OC collaterals directed to vestibular nuclei corner of granule cells at the dorsomedial edge of the VCN. raises the issue of whether OC collaterals are also directed No OC collaterals from any species were directed toward into the vestibular periphery. We found no such collaterals the granule-cell regions of the DCN-neither to the stria1 in our material. In three mice and three gerbils, a total of corner of granule cells near the dorsal acoustic stria nor to ten thin and 40 thick fibers were traced from the injection layer I1 (the pyramidal or granule cell layer). Most OC site in the spiral ganglion centrally across the inferior terminals were formed in the VCN between granule-cell vestibular ganglion and into the vestibular nerve root. None regions and the subjacent large-cell regions. A few termi- of these fibers formed branches directed into the vestibular nals were found within the granule-cell reigons and others or vestibular ganglia. One of the thick fibers formed were within the large-cell regions. The large-cell regions a branch near the inferior vestibular ganglion, but this innervated were AVCN, PVCN, and the deep DCN. branch travelled to the CN. Two other thick fibers and one In cats, four thick efferent axons produced 13 total thin fiber branched in the , but all branches were branches, with two to five branches produced per axon. directed into the auditory nerve toward the cochlea. It is They were primarily directed into the superficial granule- more difficult to address the issue of MOC fiber branching cell region of the AVCN (four out of four axons, ten of 13 to the vestibular periphery in cats, since our cat tissue was branches) and to the interstitial nucleus of the vestibular divided to process the temporal bone separately from the

‘Although separate granule-cell regions can be defined, the re- 2A reconstruction was considered complete if there was no fad- gions merge at boundaries, forming a continuous granule-cell ing of any part of the axon and each branch could be traced to a domain. terminal swelling.

BRAINSTEM BRANCHES OF OLIVOCOCHLEAR AXONS 597

0.1 rnrn f I VCN

Fig. 5. Camera-lucida reconstruction of a single thick OC axon and a and to the inferior vestibular nucleus (dashed lines). The thin axnn IS single thin OC axon in the vestibular nerve root of a mouse (modified unbranched. Large arrowheads indicate nodes of Ranvier on the thick axon. sagittal plane). The thick axon branches to the medial sheet of granule cells Inset: Low-power drawing illustrating the position of the reconstruction (dashed lines) along the medial border of the ventral cochlear nucleus (VCN) that is near the CN.

brainstem. Although it is conceivable that branches were swellings and branchlets are plentiful. In practice, we de- missed at this separation, we saw no indications that MOC fined the beginning of the terminal region as the point at fibers branched to the vestibular periphery. which there were more than five swellings or branch points Fibers of the OCB stain positively for acetylcholinester- within a 20-pm length of the branch. About 60% of all ase (AChE), and this observation has been used to explore branches appeared to be myelinated (having periodic, darkly the course of the OCB and its endings in the cochlea labeled constrictions or nodes of Ranvier) in the nonter- (Churchill and Schuknecht, ’59; Ishii and Balough, ’68) as minal region, but all appeared to be unmyelinated in the well as the CN (McDonald and Rasmussen, ’71; Martin, ’81; terminal region. Branch morphology and length differ ac- Osen et al., ’84). Our AChE-stained material from cats, cording to the region innervated (Figs. 8, 9). Branches gerbils, and mice is consistent with these earlier AChE within the vestibular nerve root in cats and to the vestibu- studies. AChE-positive fibers form a compact bundle within lar nuclei in rodents are shorter and thinner than those the vestibular nerve root. Some AChE-positive fibers enter directed to the lamina and the medial sheet of granule cells the VCN from its medial aspect, and AChE-positive fibers in the CN. Vestibular branches, like CN branches, form can be found in all regions innervated by HRP-labeled OC obvious en passant and terminal swellings in their termi- collaterals described above. Other AChE-positive fibers nal regions (Fig. lo), and the sizes of the the terminal however, are also seen in the stria1 corner of the DCN and swellings in vestibular nuclei are not obviously different to some extent throughout the superficial layer of the DCN from those in the CN. Vestibular collaterals, however, tend proper-regions where HRP-labeled OC branches have not to form fewer swellings per collateral than those directed been seen. It is thus likely that these AChE-positive fibers to the CN (Fig. 9). In cats, the branches to the CN are originate from sources other than OC efferents. longer than in rodents, because in the cat the CN is larger and the branches are directed toward the superficial gran- Terminal morphology ule-cell layer in the VCN, away from vestibular nerve root. The brainstem branches of thick OC axons in cats and One branch in cat showed more than 160 en passant and rodents can be divided into “nonterminal” regions where 50 terminal swellings-about four times the number typi- swellings are uncommon and “terminal” regions where cally found in rodents (Fig. 9). In all species, and in both 598 M.C. BROWN ET AL.

CORONAL MODIFIED SAGITTAL

Stnal

SuDerficial

Layer Medial

VCN Lamina PVCN Superficial Layer Lamina/

Vltl n. DCN laver II anterior dorsal

1 rnrn ventrolateral dorsornedial lateral +- medial + -3-posterior ventral

Fig. 6. Schematic of the granule-cell regions of the left cochlear nucleus in the strial corner and in layer ZI. VCN granule cell regions cover the in a mouse from cresyl-violet-stained sections. The same regions are illus- lateral surface of the nucleus (superficial layer) or delineate the boundary of trated as they would appear in two section planes, coronal and modified the CN from surrounding brainstem structures (medial sheet and subpedun- sagittal. The mammalian CN consists of a layered dorsal cochlear nucleus cular corner). Some granule cells (not illustrated) are found scattered (DCN) and a non-layered ventral cochlear nucleus (PVCN, AVCN), which throughout the main body of the CN. are separated by a lamina of granule cells. DCN granule cells are abundant

SUPERFICIAL TABLE 1. Percentage of Efferent Axons Forming Branches ______LAYER MSiANR MS/AVCN LAMINA VN Cochlear nucleus Vestibular nuclei

z Rodents CAT 2 8 Thin efferents 0% 0% m (0 branches/ 15 axonsj (0branched 2 axons) % Thick efferents 67% 87% so (10 branched 15 axons) (13 branches/ 15 axonsj Cats Thick efferents 100% 75% (MOC) (10 branched 4 axons) (3 branched 4 axons) L GERBIL 2 8 B so branches makes them easily distinguishable from afferent fibers (Fig. 11).OC branches are reminiscent of the branches of “group VI” axons described by Cant and Morest (’78) in Golgi stains of the CN. The thin branches of type I1 affer- ents often enter the granule-cell regions but are separable from efferent branches, because they are thinner and their endings are smaller. The smooth terminal branches of the type I axons have rounded swellings, whereas the OC Fig. 7. Graphic distribution of thick OC branches to various subdivisions branches are quite varicose in the terminal regions, and within the brainstem for axons that could be completely reconstructed. The the swellings are serrulate and more finely di~ided.~For data are from four axons in cats, seven axons in a gerbil, and eight axons in a mouse. MSIANR: MS beneath ANR; MSIAVCN: MS beneath AVCN. efferent collaterals, the entire terminal region appears to be unmyelinated, since it is uniformly and darkly labeled. the cochlear and vestibular nuclei, OC terminals were usu- ally found in neuropil and only occasionally on somata. When several fiber types are labeled, as in the CN of our %at efferent collaterals, although quite varicose, tend to form rodent preparations, the distinctive appearance of the OC smoother swellings than rodent collaterals. BRAINSTEM BRANCHES OF OLIVOCOCHLEAR AXONS 599

MEDIAL . SHEET t 3RANCH

Fig. 8. Camera-lucida tracing of thick OC branches from different axons, ule cell regions. Boxed figures illustrate collateral morphology as seen in directed to three different regions in a mouse. Top diagram of brainstem the modified sagittal plane. Arrowheads on lower tracing indicate obvious cross section indicates regions of innervation and stippling represents gran- nodes along the nonterminal region of the branch.

This distinctive pattern is reminiscent of the peripheral ciated in part with granule-cell regions. In contrast, the morphology of OC axons, where thick efferent branches neurons to the inner hair cell region, type I afferents and appear to be unmyelinated for long distances in the osseous LOC efferents terminating on the dendrites of type I affer- spiral lamina and in the before innervating ents (Smith and Rasmussen, '63; Liberman, '801, are not the outer hair cells. Type I afferent branches in CN appear obviously associated with granule-cell regions in the CN. to be myelinated to the vicinity of their endings, just as This overall regional difference in CN trajectory does not their peripheral processes are myelinated to a point within rule out the possibility of overlap in the neural networks 50-100 pm of the inner hair cells that they contact (Liber- associated with the two hair cell groups. For instance, man, '80; Brown, '87). within the main body of the CN, there is significant overlap in axonal trajectory and swelling location from type I and DISCUSSION type II afYerents, suggesting shared postsynaptic targets Branching patterns of efferent and afferent fibers mrown et al., '88a). However, further observations with the electron microscope are required to resolve this issue. Results of this study suggest an innervation plan for the The evidence that MOC neurons produce thick, myelin- CN that maintains separate pathways for the neurons in- ated axons directed to outer hair cells and that LOC neu- nervating outer hair cells and those innervating inner hair rons produce thinner, unmyelinated axons directed to inner cells (Fig. 12). The afferents contacting outer hair cells hair cell regions has been reviewed by Guinan et al. ('83). originate from type I1 spiral ganglion cells (Spoendlin, '69; The present results are consistent with this view and fur- Kiang et al., '821, and the efferents to outer hair cells are ther suggest that the formation of collaterals to the CN is provided by the thick axons of MOC neurons (Warr and common for MOC axons and is absent for LOC axons. If our Guinan, '79; Guinan et al., '83; Liberman and Brown, '86; small sample is representative, it indicates another funda- Brown, '87). Within the CN, these afferent and efferent mental difference between the MOC and LOG efferent sys- fibers to the outer hair cells have terminal swellings asso- tems. However, several other issues need to be considered. 600 M.C. BROWN ET AL. Gerbil Mouse nervating the apical turns branch near the CN, preliminary 1000 1000 data from other HRP injections made into the apical turn in mice suggest that this is not the case (A.M. Berglund, Ir personal communication). NONTERMINAL 5oo LENGTH (pm) The branching differences for MOC and LOG axons are consistent with other HRP studies of the projections of the i olivary complex to the CN. After HRP injections in the CN, I retrogradely labeled neurons were never found (Covey et al., '84;Spangler et al., '87) or rarely found (Adams, '83) in regions known to be the origins of the LOC system, whereas many labeled somata were found in medial periolivary regions (origins of the MOC system), suggesting a promi- nent descending input to the CN. However, recent autora- diographic studies of OC projection is in the gerbil come to a different conclusion (Ryan et al., '87). Following cochlear 12 incubation with 3H-D-aspartic acid, LOC (but not MOC) somata were labeled in the brainstem and labeled fibers were seen entering the CN where they terminated in the BRANCH ~ POINTS central region of the VCN, a region where no OC collaterals were seen in the present study. These data were interpreted

0 to indicate that LOC fibers branch to innervate the CN. This conflict is not easily resolved unless the thin-fiber 80 projection to the CN involves a small portion of the LOC neurons. The HRP method has the advantage that axons EN PASSANT 4o are continuously stained and can be individually recon- SWELLINGS structed, but its disadvantage is that relatively small sam- ple sizes are obtained. Thus the present data certainly do 0 not rule out the possibility that a subset of LOG neurons sends branches to the CN. Our data on the innervation of the cochlea and CN by thin fibers are most comprehensive in the rodents (mice, gerbils, and guinea pigs). The labeling studies have been most productive in smaller species, because it appears that the thin axons only transport HRP over limited distances owMSi MS/ IVN .ILLllliLAMINA MS/ IVN (Brown et al., '88a). In cats, the central projections of type AVCN ANR ANR I1 ganglion cells have been only partially reconstructed (Ryugo et al., '86; Benson and Ryugo, '871, and there are Fig. 9. Quantitative analysis of thick OC branches in gerbils and mice. Each bar graph represents an average value per collateral. Averages are virtually no published data on collaterals of LOG axons in for the following branches in gerbils: two branches to MSIAVCN, two cats. Nevertheless, the innervation plan based on data from branches to MSIANR, and seven branches to IVN (inferior vestibular nu- rodents is not contradicted by the cat material. There might cleus); and for the following branches in mice: three branches to the lamina, be considerable differences in primates, which apparently two branches to MSIANR, and six to IVN. The nonterminal length is the distance from the parent fiber to the point where the terminal region begins, lack OC collaterals to the cochlear nucleus. This situation defined as the first point where there are at least five swellings and/or may be related to the observation that the primate CN branch points within 20 pm. The terminal length represents the sums of contains few if any granule cells (Moore, '80). the lengths of all branches in the terminal region. En passant swellings were counted if their diameters were twice the diameters of the parent MOC innervation of the vestibular nuclei fibers and terminal swellings were counted if the branches forming them were longer than 3 pm. The consistent finding of OC efferent collaterals to vestib- ular nuclei suggests a relationship between certain por- tions of the auditory and . Although other First, the thin LOC axons may form fine collaterals that do investigators (Adams and Mugnaini, '87) found collaterals not transport HRP as well as the thicker collaterals of the to these regions, their sources were not identified. One MOC axons. This does not seem likely, since the thin axons potential source of such collaterals is vestibular efferent from type I1 neurons, labeled in the same injections, give axons. As conventionally defined, the fibers labeled in the off fine branches that are easily traced within the cochlear present study are auditory, not vestibular, efferents, be- nucleus (Brown et al., '88a). An additional consideration is cause 1) they innenate the organ of Corti, 2) they do not that LOG neurons may form collaterals in the brainstem in innervate the vestibular periphery, and 3) their cell bodies a region considerably medial to the cochlear nucleus. It is are located in and around auditory nuclei, in areas reported not possible to address this question with our material, to be the origin of cochlear efferents. Thus, in , since the reaction product in LOG axons faded before reach- single efferent neurons do not appear to innervate auditory ing the cell bodies. However, thin efferent fibers labeled in and vestibular endorgans, a situation clearly different from the present study certainly do not branch near the cochlear urodele amphibians, in which a common efferent system nucleus. Finally, the efferent fibers labeled in the present innervates both the and (Fritzch and study innervate predominantly the basal turn, since the Wahnschaffe, '87). HRP injections are made through the . Al- Axon degeneration studies have shown that the inferior though it might be suggested that thin efferent fibers in- vestibular nucleus receives input primarily from the UKAINSTEM BRANC'IIES OF OLIVOCOC:HIAEN1AXONY 601

f TYPE II ' AFFERENT, + ,%.f''.

\ \

and thc sncculc (Stein and (hrpenter, '67 1. Saccular projec- presence of O(: collaterals to the vestit)ular nuclei win- tions to the inferior vestibular nucleus and to the intersti- forces the idea that portions of these nuclei also have audi- tial nucleus of the vestibular nerve havc subsequentlv been torv fiinctions. demonstrated in HRP studies (Kevetter and Pkrachio, '85; Didier et al.. '87). The serves an auditorv function .l.IOC innervation of the cochlear nucleus in fish (Furukawa and Ishii, '671 and may have an aiiditoiy The mammalian (:N has been subdivided according to function in manimnls as well, at least for sounds of'moder- the projection of the auditory nerve onto the CN and the atci and high Icvcls (Cazals et al., '83).In this context, the distribution ofcc>ll types within thc (:N (Brawer et al., '74; 602 M.C. BROWN ET AL.

COCHLEAR the endings of type I neurons do not usually invade granule- COCHLEA NUCLEUS BRAINSTEM cell regions (Fekete et al., '84; Brown et al., '88a). ,. V Although observations with the light microscope are 0- r suggestive, further work with the electron microscope is n U ____ necessary to determine the postsynaptic targets for OC collaterals in the CN. The only physiological studies of the effects of efferent stimulation on responses of single CN cells have shown complex effects, sometimes inhibitory, sometimes excitatory (Starr and Wernick, '68; Comis and Whitfield, '68; Comis, '70). This complexity must reflect interaction between direct effects mediated monosynapti- AFFERENTS cally via OC collaterals or multisynaptically through inter- 0 neurons and indirect effects mediated by suppression of the responses of inner hair cells (Brown and Nuttall, '84) and type I afferents (Wiederhold and Kiang, '70). The effects of OC stimulation on CN electrophysiology should be reinves- tigated by using modern concepts of efferent anatomy and Fig. 12. Schematic diagram of the OC efferent and primary afferent cell classification of single units in the CN. Sound-driven innervation of the cochlea and CN. Thick lines indicate thick axons (> 0.7- activity in the OC collaterals may provide the CN with wm diameter) and thin lines indicate thin axons (< 0.7 qmj. In cats, the information regarding the amount of feedback the MOC LOC somata are on the margins of the LSO as drawn (Warr, '75); in rodents, neurons are providing to the hair cells, thereby calibrating the somata of the LOC neurons are within the LSO (White and Warr, '83). the nucleus for incoming messages from the auditory nerve or adjusting the gain of the afferent signal relayed to the MOC neurons themselves (Kim, '84; Guinan and Gifford, Lorente de NO, '81). The granule-cell "regions," where MOC In this context, the granule-cell regions become impor- collaterals terminate, form a rind around the main nucleus '88). tant intermediates in the MOC feedback system. (Mugnaini et al., 'Boa). These regions receive other inputs besides MOC efferents, even from nonauditory sources, in- ACKNOWLEDGMENTS cluding neurons in the cuneate nucleus and spinal trigemi- We are grateful to A.M. Berglund for her assistance in nal nucleus (Weinberg and Rustioni, '87; Itoh et al., '87). some of the experiments, to J.J. Guinan, Jr., and N.Y.S. The granule-cell regions contain granule and Golgi cells Kiang for helpful discussions, and to D.A. Learson and (Mugnaini et al., '80a), as well as other cell types (Mc- J.V. Ledwith, 111, for technical assistance. Supported by N.I.H. Donald and Rasmussen, '71j. The most common cell type, grants NS23508, NS20156, NS18339, and NS13126, N.S.F. the granule cell, shares many anatomical features with its grant BNS 8520833, and the F.V. Hunt Fellowship, Acoust- counterpart in the cerebellum, including its status as the ical Society of America. These results have been presented smallest neuron in the nucleus and the participation of its in abstract form (Brown et al., '88bj. dendrites in neural "glomeruli" clustered around a large mossy ending (Mugnaini et al., '80a,b). The synaptic con- LITERATURE CITED nections between inputs and cell types within the granule- Adam, J. (1983) Cytology of periolivary cells and the organization of their cell regions of the CN, however, are not well understood. projections. J. Comp. Neurol. 216:275-289. Our results are consistent with a previous suggestion that Adams, J.C., and E. Mugnaini (1987) Patterns of glutamate decarboxylase MOC efferents are a source of mossy endings in the VCN immunostaining in the feline cochlear nuclear complex studied with (McDonald and Rasmussen, '71). Type I1 afferent terminals silver enhancement and electron microscopy. J. Comp. Neurol. 262375- are probably too small (Brown et al., '88a), whereas the 401. largest MOC terminals are in the appropriate size range. Amesen, A.R., and K.K. Osen (1984) Flbre spectrum of the vestibulo- Some of the mossy endings have been shown to be AChE- cochlear anastomosis in the cat. Acta Otolaryngol. (Stockh.)98:225-269. Benson, T.E., and D.K. Ryugo (1987) Axons of presumptive type I1 spiral positive (McDonald and Rasmussen, '711, which is also con- ganglion neurons with granule cells ofthe cat cochlear nucleus. sistent with MOC efferents being their source. Mossy end- Soc. Neurosci. Abstr. 13:1258. ings that are not AChE-positive may also be from the OC Brawer, J.R., D.K. Morest, and E.C. Kane (1974) The neuronal architecture system since not all OC efferents stain for AChE (Vetter et of the cochlear nucleus of the cat. J. Comp. Neurol. 155:251-300. al., '86). Many of the en passant and terminal swellings Brodal, A., and 0. Pompeiano (1957) The vestibular nuclei in the cat. J. formed by MOC collaterals, however, are probably too small Anat. 91:438-454. to form mossy endings. These may correspond to the smaller Brown, M.C. (1987) Morphology of labeled efferent fibers in the guinea pig boutons, another class of AChE-positive endings studied by cochlea. J. Comp. Neurol. 26U:605-618. McDonald and Rasmussen ('71). Although mossy fibers Brown, M.C., and A.L. Nuttall (1984) Efferent control of cochlear inner hair cell responses in the guinea pig. J. Physiol, (Lond.1 354.625-646. clearly innervate granule and Golgi cells, the postsynaptic Brown, M.C., A.M. Berglund, N.Y.S. Kiang, and D.K. Ryugo(1988a)Central targets of the bouton endings are unknown. trajectories of type II spiral ganglion neurons. J. Comp. Neurol. 278:581- Type I1 afferents and MOC efferents, which converge on 590. granule-cell regions, have overlapping distributions in some Brown, M.C., M.C. Liberman, and D.K. Ryugo (1988b) Labeled collaterals areas but not in others. Endings from these two fiber types of medial olivocochlear efferents in the cochlear nucleus of cats and rodents. Presented at the Eleventh Midwinter Research Meeting of the are in close proximity in the lamina, the subpeduncular Association for Research in Otolaryngology, Clearwater, Florida. corner, and the interstitial nucleus of the . Cant, N.B., and D.K. Morest (1978) Axons from non-cochlear sources in the However, only MOC fibers appear to terminate in the me- anteroventral cochlear nucleus of the cat. A study with the rapid Golgi dial sheet and only type I1 neurons appear to end in layer method. Neuroscience 3: 1003-1029. I1 of the DCN. Light microscopic observations indicate that Cazals, Y., J.M. Aran, J.P. Erre, and A. Guilhaume (1983)Vestibular acous- BRAINSTEM BRANCHES OF OLIVOCOCHLEAR AXONS 603

tic reception in the guinea pig: A saccular function? hcta Otolaryngol. chemistry and their applications for tracing neural pathways-axonal (Stockh.) 95t211-217. transport, enzyme histochemistry and light microscopic analysis. In M.. Churchill, J.A., and H.F. Schuknecht (1959) The relationship of acetylcho- M. Mesulam (ed): Tracing Neural Connections With Horseradish Per- linesterase in the cochlea to the olivocochlear bundle. Henry Ford Hosp. oxidase. Chichester: John Wiley and Sons. pp. 1-151. Med. Bull. 7202-210. Moore, J.K. (1980) The primate cochlear nuclei: Loss of lamination as a Comis, S.D. (1970) Centrifugal inhibitory processes affecting neurones in phylogenetic process. J. Comp. Neurol. 193t609-629. the cat cochlear nucleus. J. Physiol. (Lond.) 21Ot751-760. Mugnaini, E., W.B. Warr, and K.K. Osen (1980a) Distribution and light Comis, S.D., and I.C. Whitfield (1968) Influence of centrifugal pathways on microscopic features of granule cells in the cochlear nucleus of cat, rat, unit activity in the cochlear nucleus. J:Neurophysiol. 3Ir62-68. and mouse. J. Comp. Neurol. 192:581-606. Covey, E., D.R. Jones, and J.H. Casseday (1984) Projections from the supe- Mugnaini, E., K.K. Osen, A,-L. Dahl, V.L. Friedrich, Jr., and G. Korte rior complex to the cochlear nucleus in the tree shrew. J. Comp. Neurol. (1980b) Fine structure of granule cells and related (termed 226~289-305. Golgi cells) in the cochlear nuclear complex of cat, rat, and mouse. J. Comp. Neurol. 9537-570, Didicr, A,, Y. Cazals, and C. Aurousseau (1987) Brainstem connections of the anterior and posterior parts of the saccule in the guinea pig. Acta Osen, K.K., E. Mugnaini, A.-L. Dahl, and A.H. Christiansen (1984) Histo- Otolaryngol. (Stockh.) 104t385-391. chemical localization of acetylcholinesterase in the cochlear and supe- rior olivary nuclei. A reappraisal with emphasis on the cochlear granule Fekete, D.M., E.M. Rouiller, M.C. Liberman, and D.K. Ryugo (1984) The system. Arch. Ital. Biol. 122;169-212. central projections of intracellularly labeled autditory nerve fibers in cats. J. Comp. Neurol. 229r432-450. Rasniussen, G.L. (1946) The olivary peduncle and other fiber connections of the superior olivary complex. J. Comp. Neurol. 84:141-219. Fritzsch, B., and U. Wahnschaffe (1987) Electron microscopical evidence for Rasmussen, G.L. (1953) Further observations on the efferent cochlear bun- common inner ear and lateral line efferents in urodeles. Neurosci. Lett. dle. J. Comp. Neurol. 99:61-94. 81~8-52. Rasmussen, G.L. (1960) Efferent fibers of the cochlear nerve and cochlear Furukawa, T., and Y. Ishii (1967) Neurological studies on hearing in gold- nucleus. In G.L. Rasmussen and W. Windle (eds): Neural Mechanisms fish. J. Neurophysiol. 30t1377-1403. of the Auditory and Vestibular Systlems. Sprin$ield: C.C. Thomas, pp. Ginzberg, R.D., and D.K. Morest (1983) A study of cochlear innervation in 105-115. the young cat with the Golgi method. Hear. Res. 10227-246. Rasmussen, G.L. (1967) Efferent connections of the cochlear nucleus. In Guinan, J.J., Jr., W.B. Warr, and B.E. Norris (1983) Differential olivococh- A.B. Graham (ed): Sensorineural Hearing Processes and Disorders. Bos- lear projections from lateral vs. medial zones of the superior olivary ton: Little, Brown, and Co., pp. 61-75. complex. J. Comp. Neurol. 221 :358-370. Ryan, A.F., I.R. Schwartz, R.H. Helfert, E. Keithley, and Z.-X. Wang (1987) Guinan, J.J.,Jr., and M.L. Gifford (1988) Effects of electrical stimulation of Selective retrogade labeling of lateral olivocochlear neurons in the efferent olivocochlear neurons on cat auditory-nerve fibers. 111. Tuning brainstem based on preferential uptake of 3H-D-aspartic acid in the curaves and thresholds at CF. Hear. Res. (In press). cochlea. J. Comp. Neurol. 255t606-616. Ishii, D., and K. Balough, Jr. (1968) Distribution of efferent endings in the R~~~~,D.K., L.W. Dodds, and N.Y.S. ~i~~~ (1986) morphology oftype organ of Corti. Acta Otolaryngol. (Stockh.) 66t282-288. I1 spiral ganglion cells in cats. SOC.Neurosci. Abstr. 12779. Itoh, K., H. Kamiya, A. Mitani, y. yasui, M. Takada, and N. Mizuno(I987) Smith, C.A., and G.L. Rasmussen (1963) Recent observation on the olivo- Direct projections from the and the spinal trigem- bundle. oto1.Rhinol, ~~~~~~~l.72;489-497. inal nuclei to the cochlear nuclei in the cat. Brain Res. 400:.l45-150. Spangler, K,M., N,B, Cant, C,K. Henkel, G.R, Farley, and W,B, Warr (1987) Kevetter, G.A., and A.A. Perachio (1985) Central projections of first order ~~~~~~di~~projections from the superior olivary complex to the co. vestibular neurons innervating the sacculus and posterior canal in the ,.hiear nucleus of the cat. J, camp, ~~~~~l,259;452-465. gerbil. Prog. Clin. Biol. Res. 176379-291. Spoendlin, H.H. (1969) Innervation patterns in the organ of Corti of the cat. Kiang, N.Y.S., J.M. Rho, C.C. Northrop, M.C. Liberman, and D.K. Ryugo Acts Otolaryngol. (Stockh,) 67:239-254. (1982) Hair-cell innervation by spiral ganglion cells in adult cats. Sci- Starr, A., and J.S. Wernick (1968) Olivocochlear bundle stimulation: Effects ence 217t175-177. on spontaneous and tone-evoked activities of single units in cat cochlear Kim D.O. (1984) Functional roles of the inner- and outer-hair-cell subsys- nucleus. J, ~~~~~~h~~i~~,31;54g-564, tems in the cochlea and brainstem. In (2.1. Berlin (ed): Hearing Science, Stein, B., and M.B. Carpenter (1967) Central projections of portions of the Recent Advances. San Diego: College Press, pp. 241-262. vestibular ganglion innervating specific parts of the labyrinth in the Liberman, M.C. (1980) Efferent in the inner hair cell area of the rhesus monkey, J, A~~~.120t281-318. cat cochlea: An electron microscopic study of serial sections. Hear. Res. Vetter, D.E., J.C. Adams, and E. Mugnaigni (1986) A dual efferent gabaergic 3t189-204. projection to the rat cochlea. Sac. Neurosci. Abstracts 12:779. Liberman, M.C., and M.E. Oliver (1984) Morphometry of intracellularly Warr, W.B. (1975) Olivocochlear and vestibular efferent neurons of the labeled of the auditory nerve: with functional feline brainstem: Their location, morphology, and number determined properties. J. Comp. Neurol. 223t163-176. by retrograde axonal transport and acetylcholinesterase histochemistry. Liberman, M.C., and M.C. Brown (1986) Physiology and anatomy of single J, camp, N~~~~~,161r159-182, olivocochlear neurons in the cat. Hear. Res. 24t17-36. Warr, W.B., and J.J. Guinan, Jr. (1979) Efferent innervation of the organ of Lorente de NO, R. (1981) The Primary Acoustic Nuclei. New York: Raven corti:mo separate systems. ~~~i~ R~~,173;152-155, Press. Weinberg, R.J., and A. Rustioni (1987) A cuneocochlear pathway in the rat. Martin, M.R. (1981) Acetylcholinesterase-positive fibers and cell bodies in ~~~~~~~i~~~~20~209-219, the cochlear nuclei of normal and reeler mutant mice. J. Comp. Neurol. White, J.S., and W.B. Warr (1983) The dual origins of the olivocochlear 197:153-167. bundle in the albino rat. J. Comp. Neurol. 219t203-214. McDonald, D.M., and G.L. Rasmussen (1971) Ultrastructural characteristics Wiederhold, M.L., and N.Y.S. Kiang (1970) Effects of electric stimulation of of synapt,c endings in the cochlear nucleus having acetylcholinesterase the crossed olivocochlear bundle on single auditory-nerve fibers in the activity. Brain Res. 28:1-18. cat. J. Acoust. SOC.Am. 48:950-965. Mesulam, M.-M. (1982) Principles of horseradish peroxidase neurohisto-